GenAI for Technical Hiring: Friend or Foe?

Your proverbial office watercooler might be awash with panic about candidates cheating on technical interviews using GenAI – but is the pearl-clutching justified? 

Sure, the stakes are high. That’s why there’s so much handwringing.  

Hiring the wrong person. Not hiring the right person. Wasting time and money. Wasting engineers’ time and money. Exacerbating recruiter-manager conflict. Declining slowly into irrelevance as the company struggles to hire the technical rockstars who power brilliant products and services…

But is GenAI genuinely such a threat? We don’t think so. 

Actually, we think it’s an opportunity. An opportunity to put your recruitment process under the microscope and build a more modern, more future-proof hiring function that’s better placed to hire badass devs. 

Let’s talk about that. 

How prevalent is cheating really? 

One transformation expert commented recently to Business Insider that AI in recruitment is “an arms race that is just going to keep accelerating”. 

It’s an explosive image, with talent professionals on one side of the battlefield pitted against a new generation of wily jobseekers who want to win at any cost.

But it’s not necessarily a helpful image.

Yes, AI is changing some of the ways we recruit. And it’s changing some of the ways jobseekers seek jobs. But are candidates flocking en-masse to GenAI to pull the wool over interviewers’ eyes? 

That’s inherently a hard question to answer, because candidates who are really cheating are also unlikely to crow about it. But we think not. 

This combative image of HR versus candidates belies the truth, that most candidates genuinely want to find a job where they’ll thrive. Most candidates don’t want to start a job where they can’t, any more than you want them to. Poor performance and eventual turnover aren’t good for anyone.

As our CEO Amanda Richardson puts it:

Trust that 99% of candidates believe in themselves enough to show their real skills, and the other 1% was always going to be undeterred.

Ultimately, yes, GenAI offers new possibilities for cheating. But possibilities have always existed, for the 1% who’re determined to cheat.

For example:

The fact there are new possibilities doesn’t mean cheating’s becoming more prevalent. And we don’t need to blow the problem out of proportion, in an echo-chamber of pearl-clutching. 

Recruiters need what you’ve always needed. An awareness that cheating can happen; an understanding of the warning signs; and then rolled-up sleeves to get on with hiring-as-usual. 

There’s another more pertinent question around the use of GenAI in technical interviews though, and that’s this: is using GenAI actually cheating in the first place? 

That’s a much more interesting question, because it pertains to the fundamentals of how you assess and hire the best people for the job. 

Is using GenAI during technical interviews even cheating? 

Say you’re interviewing for a job as a baker. It’s reasonable that the interviewer might ask you to bake some bread, right? 

But then imagine they refuse to let you use rising agents, even though rising agents are standard practice, because they want to see how you bake bread without artificial support.

It’s lose, lose. From their side, they aren’t actually assessing you on the skills you’ll use daily. Maybe you’re an excellent baker using rising agents but without, won’t do yourself justice. So they miss out on an excellent hire.

And from your side, how long before you abandon hiring processes like that entirely? It’s borderline offensive. You’re a serious baker, you want to work somewhere that values your skills; somewhere that doesn’t place arbitrary restrictions during recruitment just because a tiny minority of lesser bakers might take advantage. 

That’s exactly the risk with GenAI.

AI is fundamentally changing how developers work, and that’s a great thing. McKinsey research last year found that developers can complete coding tasks up to twice as fast with generative AI, for example:

Likewise, our State of Tech Hiring 2024 report found that 70% of developers think AI will help reduce their workload, and 60% would like to use more AI at work.

Of course, meeting rooms are alight with concerns about AI. Business likes certainty, not change, and anything that potentially impacts productivity is high stakes.  

But the direction of travel is towards embracing AI. Ultimately that’ll mean your developers increasingly using these tools day-to-day–so eventually, you’ll need to be hiring for those skills, just as you would any other job requirement. 

Moreover, even if GenAI isn’t yet a priority for your organization, this has an impact today – because recruitment’s fundamentally future-looking.  

That is: let’s presume you’re hiring with a view to developers staying, say, three-plus years. That means you need to hire people now with an eye on tomorrow. Building a future-ready workforce means understanding how skills needs could evolve and building them into the organization proactively. 

Which brings us back to baking. It just doesn’t make much sense, blocking candidates at interviews from using tools that they either already or almost certainly will use on the job. 

Plus, there’s a strong argument to say GenAI is an incredible acceleration opportunity for diversity, democratizing skills that have traditionally been guarded and gatekept. Improving inclusiveness and widening the talent pool are major priorities for approximately everyone. 

GenAI: friend. Right? But then we come full circle.

Because if you authorize GenAI for technical interviews, what about the candidates who will take the opportunity to cheat? And how can you build an accurate understanding of candidates’ abilities, if they’ve used AI?

Even if it’s only 1%, you still need to protect against the risk of hiring the wrong people. Or missing out on the right people, because you chose someone ahead of them who used GenAI to puff out their feathers. 

We know this stuff is a big concern. But luckily, we’re not advocating the wild, wild west. You can accommodate GenAI in a sensible, measured way that safeguards your process and protects the organization. 

How to have your cake and eat it: mitigated GenAI 

How can you harness GenAI in the recruitment process but also protect against cheating? There’s a simple answer and a longer, more nuanced answer.

The simple answer is to choose assessment and interview tools that allow candidates to use GenAI – but also have built-in functionality to discourage, detect and respond to suspicious activity. 

For instance, CoderPad has a heap of features to help mitigate cheating and prompt areas for further discussion, to help you get under the skin of a candidate’s approach. Like:

• Code playback to see how candidates wrote their code
• IDE exit detection to know if/when candidates left their test
• Copy/paste tracking to spot if code might’ve originated elsewhere
• Plagiarism detection to spot if exact code is reused
• Location tracking to check candidates are where they say they are
• Anomaly alerts for unusual candidate activity or performance 
• Candidate flagging to spot where you might need to investigate more
• Webcam proctoring and AI analysis to flag suspicious behavior 
• Question randomization to mitigate the risk of question sharing
• Question timers to prevent searching for answers
• Test performance tracking to detect unusual improvement trajectories
• AI follow-up questions to check candidates’ understanding of code

The problem of cheating on technical tests might be overblown, but it’s one that available tooling is well-equipped to tackle if support is needed. Note the ‘if’, though. ‘Anti-cheating’ functionality is a tool in your toolbox–but not everything’s a nail that needs a hammer. 

That’s the simple answer. But there’s a but, because the real answer is much more nuanced. Of course.

If you want to avoid getting into the “arms race” we spoke about earlier–that endless and combative race to the bottom–it’s less about monitoring cheating and more about building realistic, engaging hiring and assessment processes that empower candidates’ true skills to shine. 

So there’s less incentive to cheat, and both sides of the table can make an informed decision about job fit.

Prioritizing the candidate experience with a realistic, engaging recruitment process

Last year, 37.5% of organizations told Infragistics they’re struggling to find skilled developers. Good devs are in high demand. 

That means your best candidates are probably in-process with plenty of other companies. How much patience would you have for answering the same set of by-rote questions again and again before you turned to Google or AI? 

The fact is, if your process relies on copy/paste questions, you’re leaving the door open for copy/paste answers.

A recent experiment from Interview.io found that verbatim LeetCode-style questions had a 38% higher pass rate using ChatGPT than the control group not using ChatGPT, for example (53% compared to 73%). 

And worse, you’re probably driving away good people who aren’t willing to “waste time” on “useless arbitrary puzzle questions”. (Not good anyway; extra not-good given the severe developer shortage). 

If you don’t treat candidates with respect, it’s not all that surprising if they don’t respect your process.

A better way is to deliver realistic, natural, custom-to-you assessments and interviews that actually relate to the real-world job. In an engaging format that’s worth candidate’s time (while ideally demanding less internal time to manage, especially from your engineers). 

Let candidates use GenAI, if that’s how they work best. But within a platform that flags when and how they’re using it, so you can discuss with curiosity rather than condemnation. 

A recruitment process like that is much more valuable for everyone – giving candidates a better sense of whether the job’s for them and giving you a better sense of whether the candidate’s right for the job. 

And it’s miles more cheat-proof, too. Coming back to Interview.io’s experiment, customized questions had a dramatically lower pass rate using ChatGPT, of only 25%. 

Fear-based or opportunity-orientated: what’s your hiring philosophy? 

Lots of organizations that hire technical talent are settling for lackluster, generic processes – and then wondering why they struggle to hire or have high cheat rates. 

That’s not good enough anymore. Not if you hope to compete for top technical talent in an increasingly competitive environment where the bar for recruitment processes is continuously being raised. 

Choosing to integrate GenAI into your technical hiring process is really a question of your recruitment philosophy. 

• Do you see candidates as an enemy to crack down on; ready to pounce on any loophole and cheat? 

• Or do you recognize that you’re all on the same side, trying to honestly assess whether a career move is right? 

• Do you settle for a generic, boring, repetitive recruitment process that insults candidates’ intelligence and integrity? 

• Or do you care about delivering a candidate experience that’s considerate, respectful, and engaging, as well as pragmatic?

We know which side will wind-up with the strongest dev teams. And from there, the best processes, products and services, and the happiest customers and strongest bottom lines. 

That’s a recipe for high-value technical recruitment teams who are taken seriously.